Estimation Model of Rice Aboveground Dry Biomass Based on the Machine Learning and Hyperspectral Characteristic Parameters of the Canopy

Accurately estimating aboveground dry biomass (ADB) is crucial. The ADB of rice has primarily been estimated using vegetation indices with several discrete bands; nevertheless, these cannot take advantage continuous bands available hyperspectral remote sensing. This study analyzed the quantitative relationship between canopy characteristic parameters (HCPs) and rice. Twenty HCPs were used, including red edge area (SDr), blue (SDb), others. variable-screening methods involved stepwise regression (SR), a coefficient (RC), variable importance in projection (vip), random forest (RF). Stepwise partial least squares employed traditional linear as well machine learning (RF), support vector (SVM), BP artificial neural network (BPNN), an extreme machine. Whole- screening-variable models constructed to estimate at jointing, booting, heading, maturing stages across growth stages. Screening-variable include SVM based on SR (SVM-sr), RF vip (RF-vip), results show that had significant correlation containing elements region, namely SDr, SDr/SDb, (SDr − SDb)/(SDr + SDb) each stage. In addition, screening performance was better than RC RF, fewer variables screened. Moreover, region screened different Among them, SDr/SDb appeared frequently, indicating they are important. Furthermore, stage, could be well-estimated diverse modeling method found best for estimation; independent RF-vip model